Scenic Image Hierarchical Representation and Associative Memory

Abstract

Natural scenic image representation is a very active and promising research domain both in images visual perception, understand, modeling and in image retrieval. A hierarchical organization and representation approach for conceptualized scene images is proposed. Based on scenic image semantic learning, a semantic representation vector for a scenic category is formed so that three level codes hierarchy path for a scenic image is made up. In addition, in order to strengthen scenic image perception and scenic components memory, a semantic scenic associative memory approach is presented. In terms of part clues, the associative components of the scene are reminded. The experiment shows that increase the scenic elements association may enhance the recall ability.